Merge pull request #3430 from scipp/fix-group-auto-groups-str

Fix `group` with automatic groups by labels of str dtype

src/scipp/core/binning.py

@@ -704,7 +704,7 @@ def _make_groups(x, arg):
     if coord.bins is not None:
         coord = coord.copy().bins.constituents['data']
 
-    if coord.values.size == 0:
+    if 0 in coord.shape:
         unique = coord.values[0:0]
     # We are currently using np.unique to find all unique groups. This can be very slow
     # for large inputs. In many cases groups are in a bounded range of integers, and we

tests/binning_test.py

@@ -1068,3 +1068,11 @@ def test_noncontiguous_int_grouping():
     assert contiguous_idx.values.data.contiguous
 
     assert_identical(table.group(nonnontiguous_idx), table.group(contiguous_idx))
+
+
+def test_group_automatic_groups_works_with_string_coord():
+    keys = sc.array(dims=['row'], values=['a', 'b', 'c', 'a', 'b', 'c'])
+    table = sc.DataArray(sc.ones(dims=['row'], shape=[6]), coords={'key': keys})
+    assert sc.identical(
+        table.group('key').coords['key'], sc.array(dims=['key'], values=['a', 'b', 'c'])
+    )